Map端:
为来自不同表或文件的key/value对,打标签以区别不同来源的记录。然后用连接字段作为key,其余部分和新加的标志作为value,最后进行输出。
Reduce端:
在每一个分组当中将那些来源于不同文件的记录(在Map阶段已经打标志)分开,最后进行合并。
需求:
通过将关联条件作为Map输出的key,将两表满足Join条件的数据并携带数据所来源的文件信息,发往同一个ReduceTask,在Reduce中进行数据的串联。
创建商品和订单合并后的TableBean类
package com.study.mapreduce.reducejoin;
import org.apache.hadoop.io.Writable;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
public class TableBean implements Writable {
private String id;
private String pid;
private int amount;
private String name;
private String flag;
public TableBean() {
}
public String getId() {
return id;
}
public void setId(String id) {
this.id = id;
}
public String getPid() {
return pid;
}
public void setPid(String pid) {
this.pid = pid;
}
public int getAmount() {
return amount;
}
public void setAmount(int amount) {
this.amount = amount;
}
public String getName() {
return name;
}
public void setName(String name) {
this.name = name;
}
public String getFlag() {
return flag;
}
public void setFlag(String flag) {
this.flag = flag;
}
@Override
public void write(DataOutput dataOutput) throws IOException {
dataOutput.writeUTF(id);
dataOutput.writeUTF(pid);
dataOutput.writeInt(amount);
dataOutput.writeUTF(name);
dataOutput.writeUTF(flag);
}
@Override
public void readFields(DataInput dataInput) throws IOException {
this.id = dataInput.readUTF();
this.pid = dataInput.readUTF();
this.amount = dataInput.readInt();
this.name = dataInput.readUTF();
this.flag = dataInput.readUTF();
}
@Override
public String toString() {
return id + "\t" + amount + "\t" + name;
}
}
TableMapper类
package com.study.mapreduce.reducejoin;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import java.io.IOException;
public class TableMapper extends Mapper<LongWritable, Text, Text, TableBean> {
private String filename;
private Text outK = new Text();
private TableBean outV = new TableBean();
@Override
protected void setup(Context context) throws IOException, InterruptedException {
//获取对应文件名称,每一个文件只获取一次
InputSplit split = context.getInputSplit();
FileSplit fileSplit = (FileSplit) split;
filename = fileSplit.getPath().getName();
}
@Override
protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
//1 获取一行数据,转成字符串
String line = value.toString();
//2 判断是哪个文件
if(filename.contains("order"))
{
//3 切割数据
String[] split = line.split(" ");
//4 封装
outK.set(split[1]);
outV.setId(split[0]);
outV.setPid(split[1]);
outV.setAmount(Integer.parseInt(split[2]));
outV.setName("");
outV.setFlag("order");
}else{
//3 切割数据
String[] split = line.split(" ");
//4 封装
outK.set(split[0]);
outV.setId("");
outV.setPid(split[0]);
outV.setAmount(0);
outV.setName(split[1]);
outV.setFlag("pd");
}
//5 写出outK outV
context.write(outK, outV);
}
}
TableReducer类
package com.study.mapreduce.reducejoin;
import org.apache.commons.beanutils.BeanUtils;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Reducer;
import java.io.IOException;
import java.lang.reflect.InvocationTargetException;
import java.util.ArrayList;
public class TableReducer extends Reducer<Text, TableBean, TableBean, NullWritable> {
@Override
protected void reduce(Text key, Iterable<TableBean> values, Context context) throws IOException, InterruptedException {
//reduce方法是相同的key调用一次 key是pid,两个表中根据pid取order的id和amount,取pd的name
ArrayList<TableBean> orderBeans = new ArrayList<>(); //订单有多行
TableBean pdBean = new TableBean(); //只有一行
for (TableBean value : values) {
if("order".equals(value.getFlag()))
{
//创建一个临时TableBean对象接收value
//此处创建一个临时对象的原因:
//因为bean直接add是传递地址,每次循环都创建一个新的对象,赋予新的地址,再加入到集合中
TableBean tmpOrderBean = new TableBean();
try {
//工具类,把value的值暂时放到临时对象中,以免迭代中被覆盖
BeanUtils.copyProperties(tmpOrderBean,value);
} catch (IllegalAccessException e) {
e.printStackTrace();
} catch (InvocationTargetException e) {
e.printStackTrace();
}
//将临时TableBean对象添加到集合orderBeans
orderBeans.add(tmpOrderBean);
}else{
try {
BeanUtils.copyProperties(pdBean,value);
} catch (IllegalAccessException e) {
e.printStackTrace();
} catch (InvocationTargetException e) {
e.printStackTrace();
}
}
}
//遍历集合orderBeans,替换掉每个orderBean的pid为pname,然后写出
for (TableBean orderBean : orderBeans) {
orderBean.setName(pdBean.getName()); //合并
//写出修改后的orderBean对象
context.write(orderBean,NullWritable.get());
}
}
}
TableDriver类
package com.study.mapreduce.reducejoin;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import java.io.IOException;
public class TableDriver {
public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException {
//1 获取job对象
Configuration conf = new Configuration(true);
Job job = Job.getInstance(conf);
//2 关联本Driver类
job.setJarByClass(TableDriver.class);
//3 关联Mapper和Reducer
job.setMapperClass(TableMapper.class);
job.setReducerClass(TableReducer.class);
//4 设置Map端输出KV类型
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(TableBean.class);
//5 设置程序最终输出的KV类型
job.setOutputKeyClass(TableBean.class);
job.setOutputValueClass(NullWritable.class);
//6 设置程序的输入输出路径
FileInputFormat.setInputPaths(job, new Path("D:\\tableinput"));
FileOutputFormat.setOutputPath(job, new Path("D:\\tableoutput"));
//7 提交Job
boolean b = job.waitForCompletion(true);
System.exit(b ? 0 : 1);
}
}
在Reduce阶段极易产生数据倾斜,原因是大量数据在Reduce阶段进行合并处理过多的表,Map阶段负载低,资源利用率不高。
解决:在Map阶段进行数据合并操作,缓存多张表,提前处理业务逻辑,增加Map端业务,减少Reduce端数据的压力,尽可能的减少数据倾斜。
Map Join适用于一张表很小(小表缓存到内存中)、一张表很大的场景(大表用于逐行遍历)。
根据上述例子进行修改:
先将小表缓存到内存中,放入集合;逐行遍历大表的行,根据大表的关键字映射到集合中,然后取集合元素和大表元素进行拼接封装再写出。
步骤:
1.在驱动Driver中加载缓存数据,且Map阶段的join不需要Reduce阶段,所以需要设置ReduceTask数量为0。
2.在Mapper类中重写setup()和map()方法。
setup()中(处理小表):
(1)获取缓存的文件。
(2)循环读取文件的一行。
(3)切割。
(4)缓存数据到集合。
(5)关流。
map()中(处理大表以及拼接):
(1)获取一行。
(2)切割。
(3)获取pid。
(4)获取订单id和商品名称。
(5)拼接封装。
(6)写出。
MapJoinDriver类
package com.study.mapreduce.mapjoin;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import java.io.IOException;
import java.net.URI;
import java.net.URISyntaxException;
public class MapJoinDriver {
public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException, URISyntaxException {
//1 获取job对象
Configuration conf = new Configuration(true);
Job job = Job.getInstance(conf);
//2 关联本Driver类
job.setJarByClass(MapJoinDriver.class);
//3 关联Mapper
job.setMapperClass(MapJoinMapper.class);
//4 设置Map端输出KV类型
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(NullWritable.class);
//5 设置程序最终输出的KV类型
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(NullWritable.class);
//6 加载缓存数据
job.addCacheFile(new URI("file:///D:/tablecache/pd.txt"));
//7 Map端Join的逻辑不需要Reduce阶段,设置reduceTask数量为0
job.setNumReduceTasks(0);
//8 设置程序的输入输出路径
FileInputFormat.setInputPaths(job, new Path("D:\\tableinput"));
FileOutputFormat.setOutputPath(job, new Path("D:\\tableoutput"));
//9 提交Job
boolean b = job.waitForCompletion(true);
System.exit(b ? 0 : 1);
}
}
MapJoinMapper类
package com.study.mapreduce.mapjoin;
import org.apache.commons.lang3.StringUtils;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IOUtils;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.net.URI;
import java.util.HashMap;
import java.util.Map;
public class MapJoinMapper extends Mapper<LongWritable, Text, Text, NullWritable> {
//集合
private Map<String, String> pdMap = new HashMap<>();
private Text outv = new Text();
//任务开始前将pd数据缓存进pdMap
@Override
protected void setup(Context context) throws IOException, InterruptedException {
//1 通过缓存文件得到小表数据pd.txt
URI[] cacheFiles = context.getCacheFiles();
Path path = new Path(cacheFiles[0]);
//获取文件系统对象,并开流
FileSystem fs = FileSystem.get(context.getConfiguration());
FSDataInputStream fis = fs.open(path);
//通过包装流转换为reader,方便按行读取
BufferedReader reader = new BufferedReader(new InputStreamReader(fis, "UTF-8"));
//2 逐行读取,按行处理
String line;
while (StringUtils.isNotEmpty(line = reader.readLine())) {
//3 切割一行
//01 小米
String[] split = line.split(" ");
//4 缓存到集合重(pid,name)
pdMap.put(split[0], split[1]);
}
//5 关流
IOUtils.closeStream(reader);
}
@Override
protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
//1 获取一行数据,转成字符串
String line = value.toString();
//2 切割
String[] split = line.split(" ");
//3 封装
String name = pdMap.get(split[1]);
outv.set(split[0]+" "+name+" "+split[2]);
//写出
context.write(outv,NullWritable.get());
}
}